PROJECT SUMMARY Connecting genotype to phenotype remains a major challenge. Our inability to predict the fitness effects of individual mutations, particularly when the effects of mutations are non-additive, have made it difficult to connect genetic variation to phenotypes and disease traits. Changes in gene expression are thought to frequently underlie phenotypic variation. However, how mutations within regulatory regions act to dictate changes in gene expression is still not well understood. In particular, the role of additive versus non-additive (epistatic) interactions between genetic variants within regulatory regions remains largely unexplored. The proposed work will use an innovation in the CRISPR/Cas9 system (?CRISPEY?) to investigate the role of epistasis in regulatory variation and evolution. CRISPEY greatly increases the efficiency of traditional precision editing by generating a large number of potential donor DNAs in vivo using a bacterial retron element. In preliminary analyses, this method shows ~100% efficiency of precision editing with no off-target edits. The first CRISPEY scan assayed the fitness affects of 16,000 natural genetic variants differing between two strains of Saccharomyces cerevisiae (RM and BY). Strikingly, it was found that the effects of proximal promoter variants nearly always favored the same parental strain?s alleles. Reinforcement between variants in a cluster is not expected under neutral evolution, and provides evidence of widespread lineage-specific selection acting on promoter variants. Following this discovery, we will ask whether these proximal promoter variants affect fitness additively or epistatically by creating combinatorial edits of variants in each cluster. First, we will characterize general properties of regulatory epistasis. We will generate all possible combinations of 305 clusters of promoter variants (n=5,392). The fitness of each combinatorial edit will be compared to that of the predicted fitness based on individual variants to assess the extent, magnitude, and prevalence of different kinds of epistasis. Second, we ask how natural selection shapes epistasis within regulatory regions by comparing patterns observed in clusters of natural variants to a set of control variants. Next, we will ask how epistasis constrains paths available for adaptive evolution of cis-regulation by assaying all possible paths between the full BY and RM genotypes. Finally, we will ask whether epistasis for fitness results from epistasis for gene expression levels. We will use qRT-PCR to quantify gene expression levels under different combinatorial edits to assess the extent and magnitude of gene expression epistasis between natural variants. This will be the first study to conduct a genome-wide survey of epistasis between natural variants within regulatory regions. This work is critical to understanding how genetic variation translates to phenotypic variation, which is relevant for understanding the genetic basis of human disease.